Handicap The Late Money

[Franco Santiago]

Originally Posted by Dave Schwartz

... - proved to be correct, but not enough to be profitable.

I like this part of your post the most! If that ain't the truth about so many things!

[formula_2002]

evd r5 3/10 exacta
is another example where the structuring of wps and exacta pools can yield an advantage of 131%
$1 win bet returned $2.50
I structure tote board data at 4 min to post for a $1 exacta return of $3.28.
final tote board data produced a $1 return of only $2.74

		"Evangeline Downs
Race 
5

Post 
8:43

MTP 
4

$1 Exacta Pool
$8,095"	
5	OVER	pgno	bet size
		1	13
		2	22
		3	3
		4	15
		5	0
		6	2
		7	11
		8	6
		9	6
		10	14
		11	6
		12	1
		13	0
		14	0
		15	0
		16	0
			100

[Parkview_Pirate]

If you want to learn about late money and predicting final odds, play the races from Australia or Hong Kong for a while - there's plenty of late "steam" in those pools, which is not a surprise for the "North American" small pools for Australia, but is rather surprising considering the Hong Kong pools. There's some early steam too, in an attempt to drive bettors to other horses, but at least those horses don't often have the "amazing form reversals" like here in the States.

I would think the key to all this would be to find that rare horse that will win, and where the odds go UP or remain the same late in the betting. Another goal might be to identify and avoid the horses that would become underlays with the late money.

I avoid the North American tracks where late bets cut the horses' odds in half (i.e., Woodbine) as a matter of principle, though an occasional overlay is created when the steam horse fails to fire.

At the end of the day, I think tote analysis is a pretty dicey way to make a buck, especially in the U.S. where the insiders know which horse(s) is going to try, and perhaps try with a little help from the vet. It's also rather irritating to have that "slow" tote update for some pools, where the mysterious last flash often displays the winner at lower odds, even longshots lower than on-track, which is just a wee bit suspicious.

[coacht]

I was reading your post from last year and wondered if you are still using the tote board from 20 minutes to one minute to post time and if you had any modifications or updates to your post.

Thanks, coacht

[lamboguy]

i would love to know how anyone can project what the final odds will be when the last flash before the bell a horse has $8000 to win on him and crosses the wire with only $6000.

if anyone can figure that one out they will win the Nobel Prize in Economics and will outdo the Black Schoales volatility theory which did win the Nobel Prize.

[JerryBoyle]

I'm working on some analysis to eventually do exactly what the OP suggested - build a model which estimates the final price. The analysis includes the size of late changes, how "smart" the changes are, and finally if exactas and doubles can be used to help us. I'll post that analysis here when I'm done, but the punchlines are:

- the money that flows in late is smarter (we all knew this already)
- the final odds after all betting has concluded and the bets have been aggregated are the smartest you'll see (we all knew this)
- the largest increase in efficiency, where efficiency is defined as money moving generally to winning runners, is from the price you see at the start of the race to the final price. And this jump is much larger relative to going from 90 seconds out to 30s out*
- the only way to get the exact final odds would be to have a "god" model which has all inputs used by all bettors. Obviously this is impossible, so we must be comfortable with the idea that whatever model we build will not be perfect. It will be wrong, but the goal should be to reduce some measure, (squared error, r2, etc) between the price we think we're getting and the price we actually get
- there may be hope in the exacta pools for identifying where smart money is heading

This is all stuff we knew already, but I'll provide some charts/numbers later to hopefully illustrate it all.

* this is not implying that I think there is past posting. It has been covered here ad nauseam, and I'm a firm believer that there is not past posting on purpose and that any incidents of it are rare tech issues. However, if all bettors bet at the last final few seconds, it's not possible to aggregate all those bets in time for the NEXT bettor to see them and wager. Therefor, there is no bettor or team who knows the final price prior to being able to bet.

[coacht]

Thanks Jerry

I look forward to your final analysis, chart, etc.

coacht

[AskinHaskin]

Originally Posted by JerryBoyle

- the money that flows in late is smarter (we all knew this already)
- the final odds after all betting has concluded and the bets have been aggregated are the smartest you'll see (we all knew this)
- the largest increase in efficiency, where efficiency is defined as money moving generally to winning runners, is from the price you see at the start of the race to the final price. And this jump is much larger relative to going from 90 seconds out to 30s out*
- the only way to get the exact final odds would be to have a "god" model which has all inputs used by all bettors. Obviously this is impossible, so we must be comfortable with the idea that whatever model we build will not be perfect. It will be wrong, but the goal should be to reduce some measure, (squared error, r2, etc) between the price we think we're getting and the price we actually get
- there may be hope in the exacta pools for identifying where smart money is heading

... yeah, and when horse racing pulls its head out of its ass and starts putting the public on the live horses to begin with, then this stuff will revert to the generally-irrelevant category it used to occupy.


You're the only gamble in town which can do that, so WTF wouldn't you...?

[AltonKelsey]

I'm sincerely curious.



Anyone here have a model that wins even AFTER knowing the final odds ?

[JerryBoyle]

Originally Posted by AltonKelsey

I'm sincerely curious.



Anyone here have a model that wins even AFTER knowing the final odds ?

I have a model that WOULD win if I knew the final odds when I made a bet. Unfortunately, as we all know, the odds swings between at the gate and after start are so large that it's hard to know if it'll actually win .

[clicknow]

Originally Posted by lamboguy

i would love to know how anyone can project what the final odds will be when the last flash before the bell a horse has $8000 to win on him and crosses the wire with only $6000. .

Cancelled bets at last minute? A way to manipulate pools?

[aaron]

Originally Posted by lamboguy

i would love to know how anyone can project what the final odds will be when the last flash before the bell a horse has $8000 to win on him and crosses the wire with only $6000.

if anyone can figure that one out they will win the Nobel Prize in Economics and will outdo the Black Schoales volatility theory which did win the Nobel Prize.

Interesting topic. If you write down the DD will pays, more often than not, the odds will be reflected in the final odds of the race. Not unusual, for a horse who is 9/2 with a no time on the clock to go off 9/5 if he/she was 1st or 2nd choice in the double. There are exceptions,but it does happen more often than not. Last week at Aqueduct a first time starter was bet to favorite or 2nd choice in the doubles, but paid the same win price as the double. Strange result.

[clicknow]

Originally Posted by coacht

I was reading your post from last year and wondered if you are still using the tote board from 20 minutes to one minute to post time and if you had any modifications or updates to your post.

1 MTP is way too long. If you go back and read the Firenze Fire topic in the Dwyer the late money (over $120K) came in something like the last 28 seconds.

Not that it matters, since as a bettor with a normal internet connection, without some special help, (technology-wise) or a very good friend at the ticket counter who can press a button at 5 sec to post, you will not be able to actually *use* that information, because by the time you perceive a big odds drop due to a large bet or bets, it's way too late for you to even make a wager.

So, while I admire Jerry and others for toying w/the idea of writing some neat programs, the bottom line is that unless you are directly wired in somewhere, your wager is not going to make the cutoff.

Originally Posted by AltonKelsey

I'm sincerely curious.
Anyone here have a model that wins even AFTER knowing the final odds ?

Apparently, you dont have to even know the odds. All you have to know is that a horse is very likely to win a race that even if their M/L is 6-1, it's a sure enough bet ........only way anyone would think that is IF they knew horse had been given some kinda rocket fuel ........... because you can't tell me that even a really good honest trainer *know* their horse is going to win a race. Horses are just not that predictable, and even if they were, stuff can "happen" during the race that makes their win impossible.

But the CRW teams and whales make enough in rebates to cover their losses, so it doesn't matter in the long run. YOUR $60K wager on such a horse isn't likely to turn out as well if he does not win, since you don't have that kind of insurance

[azeri98]

I also like to see early money, for example at Gulfstream on Saturday in race 4 the opened at 7 to 1 closed at 13 to 1 m/l was 20 to 1 race 9 the opened at 8 to 1 closed at 13 to 1, both won and Sunday race 5 the opened at 9/5 and closed at 12 to 1 and finished 2nd. It happens fairly often.

[Lemon Drop Husker]

Originally Posted by aaron

Interesting topic. If you write down the DD will pays, more often than not, the odds will be reflected in the final odds of the race. Not unusual, for a horse who is 9/2 with a no time on the clock to go off 9/5 if he/she was 1st or 2nd choice in the double. There are exceptions,but it does happen more often than not. Last week at Aqueduct a first time starter was bet to favorite or 2nd choice in the doubles, but paid the same win price as the double. Strange result.

Sharp post.


Will Pays always have to be watched. Even in the Pick 3, 4, and 5 pay outs.